Created attachment 1484044 [details] different disk naming Description of problem: Version-Release number of selected component (if applicable): 5.10.0.15 How reproducible: Always Steps to Reproduce: 1. Boot appliance instance in ec2 env with newer instance types(t3,c5,c5d,m5d) 2. Try to create internal database 3. Actual results: Internal database cannot be created because volume partitions have different naming than older instances and appliance_console cannot find these partitions. Expected results: Additional info:
Created attachment 1484045 [details] database creation error
Could you elaborate on, or link to some documentation about, the differences in partitions or device naming that you think might be causing this issue? It looks like the paths we search for attached devices are these: '/dev/[vhs]d[a-z]', '/dev/xvd[a-z]' That expands to the following formats: vd*, hd*, sd*, xvd* This is defined here: https://github.com/ManageIQ/linux_admin/blob/master/lib/linux_admin/disk.rb#L12 If you can either provide an environment where this issue exists or explain the new format, then a fix should be fairly straight forward.
Ah, I just saw the other attachment it looks like the devices are named using the pattern `nvme[1-9]n1` maybe? Some documentation about this naming scheme would still be helpful.
Nick, there is documentation for device naming in AWS: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/device_naming.html
I don't have the environment to reproduce this so could you check if something like `lsblk -d -oNAME -n` would detect these? Maybe we can move to that as a more generic solution rather than pre-defined file names in /dev.
https://github.com/ManageIQ/linux_admin/pull/203
Matouš, I created a patch that should fix this, can you test this out on one of the problematic instances? https://patch-diff.githubusercontent.com/raw/ManageIQ/linux_admin/pull/203.patch
New commit detected on ManageIQ/linux_admin/master: https://github.com/ManageIQ/linux_admin/commit/eaff9fccfb16dd111a4bbf45d561a1afd09ad9c4 commit eaff9fccfb16dd111a4bbf45d561a1afd09ad9c4 Author: Nick Carboni <ncarboni> AuthorDate: Thu Sep 27 10:25:17 2018 -0400 Commit: Nick Carboni <ncarboni> CommitDate: Thu Sep 27 10:25:17 2018 -0400 Use lsblk to detect local disks This will remove the need to track all the different disk naming schemes using a regex. Specifically it will allow us to find nvme devices which are used in some AWS instance types https://bugzilla.redhat.com/show_bug.cgi?id=1629853 lib/linux_admin/disk.rb | 3 +- spec/disk_spec.rb | 6 +- 2 files changed, 6 insertions(+), 3 deletions(-)
New commit detected on ManageIQ/manageiq-appliance_console/master: https://github.com/ManageIQ/manageiq-appliance_console/commit/205e712f6eda97a50f4baf2eac740a1a84c7450b commit 205e712f6eda97a50f4baf2eac740a1a84c7450b Author: Nick Carboni <ncarboni> AuthorDate: Fri Sep 28 15:38:41 2018 -0400 Commit: Nick Carboni <ncarboni> CommitDate: Fri Sep 28 15:38:41 2018 -0400 Update the minimum linux_admin version This will include the fix for detecting disk devices for use when creating the local database https://bugzilla.redhat.com/show_bug.cgi?id=1629853 manageiq-appliance_console.gemspec | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Nick, I tried the patch, but it parses volumes incorrectly as it tells pvcreate to use different disk than the one which exists. Screenshot attached.
Created attachment 1490844 [details] console log + lsblk
https://github.com/ManageIQ/linux_admin/pull/205
New commit detected on ManageIQ/linux_admin/master: https://github.com/ManageIQ/linux_admin/commit/ec2ba4a0d56a58d35377fb6cf78ff7e37030ad00 commit ec2ba4a0d56a58d35377fb6cf78ff7e37030ad00 Author: Nick Carboni <ncarboni> AuthorDate: Fri May 10 15:00:10 2019 -0400 Commit: Nick Carboni <ncarboni> CommitDate: Fri May 10 15:00:10 2019 -0400 Support nvme device partition naming This commit collects the disk model from the parted output and uses that to change how we construct the partition path. Right now we keep the previous behavior of <device_path><partition_id> in all cases except nvme when we use <device_path>p<partition_id>. Previously, we would list a path which didn't exist for nvme devices. https://bugzilla.redhat.com/show_bug.cgi?id=1629853 lib/linux_admin/disk.rb | 20 +- lib/linux_admin/partition.rb | 2 +- spec/disk_spec.rb | 135 +- 3 files changed, 117 insertions(+), 40 deletions(-)
https://github.com/ManageIQ/manageiq-appliance_console/pull/90
New commit detected on ManageIQ/manageiq-appliance_console/master: https://github.com/ManageIQ/manageiq-appliance_console/commit/b7cb2f9f0962a7d7f2306de396b5814d12794a01 commit b7cb2f9f0962a7d7f2306de396b5814d12794a01 Author: Nick Carboni <ncarboni> AuthorDate: Tue May 14 17:06:19 2019 -0400 Commit: Nick Carboni <ncarboni> CommitDate: Tue May 14 17:06:19 2019 -0400 Pull in version 1.2.3 of linux admin for nvme device fix https://bugzilla.redhat.com/show_bug.cgi?id=1629853 manageiq-appliance_console.gemspec | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
https://github.com/ManageIQ/manageiq-appliance/pull/241
New commit detected on ManageIQ/manageiq-appliance/master: https://github.com/ManageIQ/manageiq-appliance/commit/75ce77ffa125ab85aa493def3a6804eddd46968a commit 75ce77ffa125ab85aa493def3a6804eddd46968a Author: Nick Carboni <ncarboni> AuthorDate: Mon Jun 10 16:20:29 2019 -0400 Commit: Nick Carboni <ncarboni> CommitDate: Mon Jun 10 16:20:29 2019 -0400 Bump console version to 5.0.0 Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1629853 Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1670150 manageiq-appliance-dependencies.rb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
Closing because it cannot be verified as we don't support EC2 images in 5.11.
reverting back to ON_QA for re-validation due to https://bugzilla.redhat.com/show_bug.cgi?id=1671909
Verified in 5.11.2.0. Database can be created in appliance_console in EC2 instance with instance types like c5d. Appliance_console parses nvme drives correctly.